Chevron Sales Prediction

Inspiration

Every year, millions of fresh food go to waste leading to lots of money lost. Almost 131 billion dollars have been lost due to food waste, and for several products, gas stations and supermarkets waste almost 10% of these food items.

What it does

I have created a machine learning model that is able to predict how many hot dogs are going to be sold ahead of time using the Random Forest Classifier.

How we built it

• I used python, sci-kit learn, pandas, numpy, seaborn, First we prepared the data by using the pandas library. After reading the data, we chose to take out any null values in case. We then converted NAN's to usable data forms Lastly we converted the data in boolean to numerical form After using several ML classifiers, in the end we chose to use Random Forest Classifier which had an accuracy of 99%+ ## Challenges we ran into --Feature Selection --Not enough time to take City and State into consideration so we dropped it from the dataframe ## Accomplishments that we're proud of
• Easy to understand visualizations
• First ever datathon
• Created a project with 99%+ accuracy ## What we learned
• learned how to use a variety of models and data visualization techniques.
• How to find the best classifier. ## What's next for Chevron Sales Prediction For the random forest classifier, I did not factor geographic location. Would like to convert geographic location into a usable form of data. Covid 19 is affecting sales in a major way. I would like to analyze the way Covid has been affecting gas stations recently and make a more well rounded specialized model for the effects Covid has caused on sales. Try using this model on other Chevron products that are also thrown away every day, and see if good results are produced.

Built With

• python